llvm/tools/verify-uselistorder/verify-uselistorder.cpp

   1 //===- verify-uselistorder.cpp - The LLVM Modular Optimizer ---------------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8 //
   9 // Verify that use-list order can be serialized correctly.  After reading the
  10 // provided IR, this tool shuffles the use-lists and then writes and reads to a
  11 // separate Module whose use-list orders are compared to the original.
  12 //
  13 // The shuffles are deterministic, but guarantee that use-lists will change.
  14 // The algorithm per iteration is as follows:
  15 //
  16 //  1. Seed the random number generator.  The seed is different for each
  17 //     shuffle.  Shuffle 0 uses default+0, shuffle 1 uses default+1, and so on.
  18 //
  19 //  2. Visit every Value in a deterministic order.
  20 //
  21 //  3. Assign a random number to each Use in the Value's use-list in order.
  22 //
  23 //  4. If the numbers are already in order, reassign numbers until they aren't.
  24 //
  25 //  5. Sort the use-list using Value::sortUseList(), which is a stable sort.
  26 //
  27 //===----------------------------------------------------------------------===//
  28
  29 #include "llvm/ADT/DenseMap.h"
  30 #include "llvm/ADT/DenseSet.h"
  31 #include "llvm/AsmParser/Parser.h"
  32 #include "llvm/Bitcode/BitcodeReader.h"
  33 #include "llvm/Bitcode/BitcodeWriter.h"
  34 #include "llvm/IR/LLVMContext.h"
  35 #include "llvm/IR/Module.h"
  36 #include "llvm/IR/UseListOrder.h"
  37 #include "llvm/IR/Verifier.h"
  38 #include "llvm/IRReader/IRReader.h"
  39 #include "llvm/Support/CommandLine.h"
  40 #include "llvm/Support/Debug.h"
  41 #include "llvm/Support/ErrorHandling.h"
  42 #include "llvm/Support/FileSystem.h"
  43 #include "llvm/Support/FileUtilities.h"
  44 #include "llvm/Support/InitLLVM.h"
  45 #include "llvm/Support/MemoryBuffer.h"
  46 #include "llvm/Support/SourceMgr.h"
  47 #include "llvm/Support/SystemUtils.h"
  48 #include "llvm/Support/raw_ostream.h"
  49 #include <random>
  50 #include <vector>
  51
  52 using namespace llvm;
  53
  54 #define DEBUG_TYPE "uselistorder"
  55
  56 static cl::opt<std::string> InputFilename(cl::Positional,
  57                                           cl::desc("<input bitcode file>"),
  58                                           cl::init("-"),
  59                                           cl::value_desc("filename"));
  60
  61 static cl::opt<bool> SaveTemps("save-temps", cl::desc("Save temp files"),
  62                                cl::init(false));
  63
  64 static cl::opt<unsigned>
  65     NumShuffles("num-shuffles",
  66                 cl::desc("Number of times to shuffle and verify use-lists"),
  67                 cl::init(1));
  68
  69 namespace {
  70
  71 struct TempFile {
  72   std::string Filename;
  73   FileRemover Remover;
  74   bool init(const std::string &Ext);
  75   bool writeBitcode(const Module &M) const;
  76   bool writeAssembly(const Module &M) const;
  77   std::unique_ptr<Module> readBitcode(LLVMContext &Context) const;
  78   std::unique_ptr<Module> readAssembly(LLVMContext &Context) const;
  79 };
  80
  81 struct ValueMapping {
  82   DenseMap<const Value *, unsigned> IDs;
  83   std::vector<const Value *> Values;
  84
  85   /// Construct a value mapping for module.
  86   ///
  87   /// Creates mapping from every value in \c M to an ID.  This mapping includes
  88   /// un-referencable values.
  89   ///
  90   /// Every \a Value that gets serialized in some way should be represented
  91   /// here.  The order needs to be deterministic, but it's unnecessary to match
  92   /// the value-ids in the bitcode writer.
  93   ///
  94   /// All constants that are referenced by other values are included in the
  95   /// mapping, but others -- which wouldn't be serialized -- are not.
  96   ValueMapping(const Module &M);
  97
  98   /// Map a value.
  99   ///
 100   /// Maps a value.  If it's a constant, maps all of its operands first.
 101   void map(const Value *V);
 102   unsigned lookup(const Value *V) const { return IDs.lookup(V); }
 103 };
 104
 105 } // end namespace
 106
 107 bool TempFile::init(const std::string &Ext) {
 108   SmallVector<char, 64> Vector;
 109   LLVM_DEBUG(dbgs() << " - create-temp-file\n");
 110   if (auto EC = sys::fs::createTemporaryFile("uselistorder", Ext, Vector)) {
 111     errs() << "verify-uselistorder: error: " << EC.message() << "\n";
 112     return true;
 113   }
 114   assert(!Vector.empty());
 115
 116   Filename.assign(Vector.data(), Vector.data() + Vector.size());
 117   Remover.setFile(Filename, !SaveTemps);
 118   if (SaveTemps)
 119     outs() << " - filename = " << Filename << "\n";
 120   return false;
 121 }
 122
 123 bool TempFile::writeBitcode(const Module &M) const {
 124   LLVM_DEBUG(dbgs() << " - write bitcode\n");
 125   std::error_code EC;
 126   raw_fd_ostream OS(Filename, EC, sys::fs::OF_None);
 127   if (EC) {
 128     errs() << "verify-uselistorder: error: " << EC.message() << "\n";
 129     return true;
 130   }
 131
 132   WriteBitcodeToFile(M, OS, /* ShouldPreserveUseListOrder */ true);
 133   return false;
 134 }
 135
 136 bool TempFile::writeAssembly(const Module &M) const {
 137   LLVM_DEBUG(dbgs() << " - write assembly\n");
 138   std::error_code EC;
 139   raw_fd_ostream OS(Filename, EC, sys::fs::OF_TextWithCRLF);
 140   if (EC) {
 141     errs() << "verify-uselistorder: error: " << EC.message() << "\n";
 142     return true;
 143   }
 144
 145   M.print(OS, nullptr, /* ShouldPreserveUseListOrder */ true);
 146   return false;
 147 }
 148
 149 std::unique_ptr<Module> TempFile::readBitcode(LLVMContext &Context) const {
 150   LLVM_DEBUG(dbgs() << " - read bitcode\n");
 151   ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOr =
 152       MemoryBuffer::getFile(Filename);
 153   if (!BufferOr) {
 154     errs() << "verify-uselistorder: error: " << BufferOr.getError().message()
 155            << "\n";
 156     return nullptr;
 157   }
 158
 159   MemoryBuffer *Buffer = BufferOr.get().get();
 160   Expected<std::unique_ptr<Module>> ModuleOr =
 161       parseBitcodeFile(Buffer->getMemBufferRef(), Context);
 162   if (!ModuleOr) {
 163     logAllUnhandledErrors(ModuleOr.takeError(), errs(),
 164                           "verify-uselistorder: error: ");
 165     return nullptr;
 166   }
 167   return std::move(ModuleOr.get());
 168 }
 169
 170 std::unique_ptr<Module> TempFile::readAssembly(LLVMContext &Context) const {
 171   LLVM_DEBUG(dbgs() << " - read assembly\n");
 172   SMDiagnostic Err;
 173   std::unique_ptr<Module> M = parseAssemblyFile(Filename, Err, Context);
 174   if (!M.get())
 175     Err.print("verify-uselistorder", errs());
 176   return M;
 177 }
 178
 179 ValueMapping::ValueMapping(const Module &M) {
 180   // Every value should be mapped, including things like void instructions and
 181   // basic blocks that are kept out of the ValueEnumerator.
 182   //
 183   // The current mapping order makes it easier to debug the tables.  It happens
 184   // to be similar to the ID mapping when writing ValueEnumerator, but they
 185   // aren't (and needn't be) in sync.
 186
 187   // Globals.
 188   for (const GlobalVariable &G : M.globals())
 189     map(&G);
 190   for (const GlobalAlias &A : M.aliases())
 191     map(&A);
 192   for (const GlobalIFunc &IF : M.ifuncs())
 193     map(&IF);
 194   for (const Function &F : M)
 195     map(&F);
 196
 197   // Constants used by globals.
 198   for (const GlobalVariable &G : M.globals())
 199     if (G.hasInitializer())
 200       map(G.getInitializer());
 201   for (const GlobalAlias &A : M.aliases())
 202     map(A.getAliasee());
 203   for (const GlobalIFunc &IF : M.ifuncs())
 204     map(IF.getResolver());
 205   for (const Function &F : M)
 206     for (Value *Op : F.operands())
 207       map(Op);
 208
 209   // Function bodies.
 210   for (const Function &F : M) {
 211     for (const Argument &A : F.args())
 212       map(&A);
 213     for (const BasicBlock &BB : F)
 214       map(&BB);
 215     for (const BasicBlock &BB : F)
 216       for (const Instruction &I : BB)
 217         map(&I);
 218
 219     // Constants used by instructions.
 220     for (const BasicBlock &BB : F)
 221       for (const Instruction &I : BB)
 222         for (const Value *Op : I.operands()) {
 223           // Look through a metadata wrapper.
 224           if (const auto *MAV = dyn_cast<MetadataAsValue>(Op))
 225             if (const auto *VAM = dyn_cast<ValueAsMetadata>(MAV->getMetadata()))
 226               Op = VAM->getValue();
 227
 228           if ((isa<Constant>(Op) && !isa<GlobalValue>(*Op)) ||
 229               isa<InlineAsm>(Op))
 230             map(Op);
 231         }
 232   }
 233 }
 234
 235 void ValueMapping::map(const Value *V) {
 236   if (IDs.lookup(V))
 237     return;
 238
 239   if (auto *C = dyn_cast<Constant>(V))
 240     if (!isa<GlobalValue>(C))
 241       for (const Value *Op : C->operands())
 242         map(Op);
 243
 244   Values.push_back(V);
 245   IDs[V] = Values.size();
 246 }
 247
 248 #ifndef NDEBUG
 249 static void dumpMapping(const ValueMapping &VM) {
 250   dbgs() << "value-mapping (size = " << VM.Values.size() << "):\n";
 251   for (unsigned I = 0, E = VM.Values.size(); I != E; ++I) {
 252     dbgs() << " - id = " << I << ", value = ";
 253     VM.Values[I]->dump();
 254   }
 255 }
 256
 257 static void debugValue(const ValueMapping &M, unsigned I, StringRef Desc) {
 258   const Value *V = M.Values[I];
 259   dbgs() << " - " << Desc << " value = ";
 260   V->dump();
 261   for (const Use &U : V->uses()) {
 262     dbgs() << "   => use: op = " << U.getOperandNo()
 263            << ", user-id = " << M.IDs.lookup(U.getUser()) << ", user = ";
 264     U.getUser()->dump();
 265   }
 266 }
 267
 268 static void debugUserMismatch(const ValueMapping &L, const ValueMapping &R,
 269                               unsigned I) {
 270   dbgs() << " - fail: user mismatch: ID = " << I << "\n";
 271   debugValue(L, I, "LHS");
 272   debugValue(R, I, "RHS");
 273
 274   dbgs() << "\nlhs-";
 275   dumpMapping(L);
 276   dbgs() << "\nrhs-";
 277   dumpMapping(R);
 278 }
 279
 280 static void debugSizeMismatch(const ValueMapping &L, const ValueMapping &R) {
 281   dbgs() << " - fail: map size: " << L.Values.size()
 282          << " != " << R.Values.size() << "\n";
 283   dbgs() << "\nlhs-";
 284   dumpMapping(L);
 285   dbgs() << "\nrhs-";
 286   dumpMapping(R);
 287 }
 288 #endif
 289
 290 static bool matches(const ValueMapping &LM, const ValueMapping &RM) {
 291   LLVM_DEBUG(dbgs() << "compare value maps\n");
 292   if (LM.Values.size() != RM.Values.size()) {
 293     LLVM_DEBUG(debugSizeMismatch(LM, RM));
 294     return false;
 295   }
 296
 297   // This mapping doesn't include dangling constant users, since those don't
 298   // get serialized.  However, checking if users are constant and calling
 299   // isConstantUsed() on every one is very expensive.  Instead, just check if
 300   // the user is mapped.
 301   auto skipUnmappedUsers =
 302       [&](Value::const_use_iterator &U, Value::const_use_iterator E,
 303           const ValueMapping &M) {
 304     while (U != E && !M.lookup(U->getUser()))
 305       ++U;
 306   };
 307
 308   // Iterate through all values, and check that both mappings have the same
 309   // users.
 310   for (unsigned I = 0, E = LM.Values.size(); I != E; ++I) {
 311     const Value *L = LM.Values[I];
 312     const Value *R = RM.Values[I];
 313     auto LU = L->use_begin(), LE = L->use_end();
 314     auto RU = R->use_begin(), RE = R->use_end();
 315     skipUnmappedUsers(LU, LE, LM);
 316     skipUnmappedUsers(RU, RE, RM);
 317
 318     while (LU != LE) {
 319       if (RU == RE) {
 320         LLVM_DEBUG(debugUserMismatch(LM, RM, I));
 321         return false;
 322       }
 323       if (LM.lookup(LU->getUser()) != RM.lookup(RU->getUser())) {
 324         LLVM_DEBUG(debugUserMismatch(LM, RM, I));
 325         return false;
 326       }
 327       if (LU->getOperandNo() != RU->getOperandNo()) {
 328         LLVM_DEBUG(debugUserMismatch(LM, RM, I));
 329         return false;
 330       }
 331       skipUnmappedUsers(++LU, LE, LM);
 332       skipUnmappedUsers(++RU, RE, RM);
 333     }
 334     if (RU != RE) {
 335       LLVM_DEBUG(debugUserMismatch(LM, RM, I));
 336       return false;
 337     }
 338   }
 339
 340   return true;
 341 }
 342
 343 static void verifyAfterRoundTrip(const Module &M,
 344                                  std::unique_ptr<Module> OtherM) {
 345   if (!OtherM)
 346     report_fatal_error("parsing failed");
 347   if (verifyModule(*OtherM, &errs()))
 348     report_fatal_error("verification failed");
 349   if (!matches(ValueMapping(M), ValueMapping(*OtherM)))
 350     report_fatal_error("use-list order changed");
 351 }
 352
 353 static void verifyBitcodeUseListOrder(const Module &M) {
 354   TempFile F;
 355   if (F.init("bc"))
 356     report_fatal_error("failed to initialize bitcode file");
 357
 358   if (F.writeBitcode(M))
 359     report_fatal_error("failed to write bitcode");
 360
 361   LLVMContext Context;
 362   verifyAfterRoundTrip(M, F.readBitcode(Context));
 363 }
 364
 365 static void verifyAssemblyUseListOrder(const Module &M) {
 366   TempFile F;
 367   if (F.init("ll"))
 368     report_fatal_error("failed to initialize assembly file");
 369
 370   if (F.writeAssembly(M))
 371     report_fatal_error("failed to write assembly");
 372
 373   LLVMContext Context;
 374   verifyAfterRoundTrip(M, F.readAssembly(Context));
 375 }
 376
 377 static void verifyUseListOrder(const Module &M) {
 378   outs() << "verify bitcode\n";
 379   verifyBitcodeUseListOrder(M);
 380   outs() << "verify assembly\n";
 381   verifyAssemblyUseListOrder(M);
 382 }
 383
 384 static void shuffleValueUseLists(Value *V, std::minstd_rand0 &Gen,
 385                                  DenseSet<Value *> &Seen) {
 386   if (!Seen.insert(V).second)
 387     return;
 388
 389   if (auto *C = dyn_cast<Constant>(V))
 390     if (!isa<GlobalValue>(C))
 391       for (Value *Op : C->operands())
 392         shuffleValueUseLists(Op, Gen, Seen);
 393
 394   if (V->use_empty() || std::next(V->use_begin()) == V->use_end())
 395     // Nothing to shuffle for 0 or 1 users.
 396     return;
 397
 398   // Generate random numbers between 10 and 99, which will line up nicely in
 399   // debug output.  We're not worried about collisons here.
 400   LLVM_DEBUG(dbgs() << "V = "; V->dump());
 401   std::uniform_int_distribution<short> Dist(10, 99);
 402   SmallDenseMap<const Use *, short, 16> Order;
 403   auto compareUses =
 404       [&Order](const Use &L, const Use &R) { return Order[&L] < Order[&R]; };
 405   do {
 406     for (const Use &U : V->uses()) {
 407       auto I = Dist(Gen);
 408       Order[&U] = I;
 409       LLVM_DEBUG(dbgs() << " - order: " << I << ", op = " << U.getOperandNo()
 410                         << ", U = ";
 411                  U.getUser()->dump());
 412     }
 413   } while (std::is_sorted(V->use_begin(), V->use_end(), compareUses));
 414
 415   LLVM_DEBUG(dbgs() << " => shuffle\n");
 416   V->sortUseList(compareUses);
 417
 418   LLVM_DEBUG({
 419     for (const Use &U : V->uses()) {
 420       dbgs() << " - order: " << Order.lookup(&U)
 421              << ", op = " << U.getOperandNo() << ", U = ";
 422       U.getUser()->dump();
 423     }
 424   });
 425 }
 426
 427 static void reverseValueUseLists(Value *V, DenseSet<Value *> &Seen) {
 428   if (!Seen.insert(V).second)
 429     return;
 430
 431   if (auto *C = dyn_cast<Constant>(V))
 432     if (!isa<GlobalValue>(C))
 433       for (Value *Op : C->operands())
 434         reverseValueUseLists(Op, Seen);
 435
 436   if (V->use_empty() || std::next(V->use_begin()) == V->use_end())
 437     // Nothing to shuffle for 0 or 1 users.
 438     return;
 439
 440   LLVM_DEBUG({
 441     dbgs() << "V = ";
 442     V->dump();
 443     for (const Use &U : V->uses()) {
 444       dbgs() << " - order: op = " << U.getOperandNo() << ", U = ";
 445       U.getUser()->dump();
 446     }
 447     dbgs() << " => reverse\n";
 448   });
 449
 450   V->reverseUseList();
 451
 452   LLVM_DEBUG({
 453     for (const Use &U : V->uses()) {
 454       dbgs() << " - order: op = " << U.getOperandNo() << ", U = ";
 455       U.getUser()->dump();
 456     }
 457   });
 458 }
 459
 460 template <class Changer>
 461 static void changeUseLists(Module &M, Changer changeValueUseList) {
 462   // Visit every value that would be serialized to an IR file.
 463   //
 464   // Globals.
 465   for (GlobalVariable &G : M.globals())
 466     changeValueUseList(&G);
 467   for (GlobalAlias &A : M.aliases())
 468     changeValueUseList(&A);
 469   for (GlobalIFunc &IF : M.ifuncs())
 470     changeValueUseList(&IF);
 471   for (Function &F : M)
 472     changeValueUseList(&F);
 473
 474   // Constants used by globals.
 475   for (GlobalVariable &G : M.globals())
 476     if (G.hasInitializer())
 477       changeValueUseList(G.getInitializer());
 478   for (GlobalAlias &A : M.aliases())
 479     changeValueUseList(A.getAliasee());
 480   for (GlobalIFunc &IF : M.ifuncs())
 481     changeValueUseList(IF.getResolver());
 482   for (Function &F : M)
 483     for (Value *Op : F.operands())
 484       changeValueUseList(Op);
 485
 486   // Function bodies.
 487   for (Function &F : M) {
 488     for (Argument &A : F.args())
 489       changeValueUseList(&A);
 490     for (BasicBlock &BB : F)
 491       changeValueUseList(&BB);
 492     for (BasicBlock &BB : F)
 493       for (Instruction &I : BB)
 494         changeValueUseList(&I);
 495
 496     // Constants used by instructions.
 497     for (BasicBlock &BB : F)
 498       for (Instruction &I : BB)
 499         for (Value *Op : I.operands()) {
 500           // Look through a metadata wrapper.
 501           if (auto *MAV = dyn_cast<MetadataAsValue>(Op))
 502             if (auto *VAM = dyn_cast<ValueAsMetadata>(MAV->getMetadata()))
 503               Op = VAM->getValue();
 504           if ((isa<Constant>(Op) && !isa<GlobalValue>(*Op)) ||
 505               isa<InlineAsm>(Op))
 506             changeValueUseList(Op);
 507         }
 508   }
 509
 510   if (verifyModule(M, &errs()))
 511     report_fatal_error("verification failed");
 512 }
 513
 514 static void shuffleUseLists(Module &M, unsigned SeedOffset) {
 515   std::minstd_rand0 Gen(std::minstd_rand0::default_seed + SeedOffset);
 516   DenseSet<Value *> Seen;
 517   changeUseLists(M, [&](Value *V) { shuffleValueUseLists(V, Gen, Seen); });
 518   LLVM_DEBUG(dbgs() << "\n");
 519 }
 520
 521 static void reverseUseLists(Module &M) {
 522   DenseSet<Value *> Seen;
 523   changeUseLists(M, [&](Value *V) { reverseValueUseLists(V, Seen); });
 524   LLVM_DEBUG(dbgs() << "\n");
 525 }
 526
 527 int main(int argc, char **argv) {
 528   InitLLVM X(argc, argv);
 529
 530   // Enable debug stream buffering.
 531   EnableDebugBuffering = true;
 532
 533   cl::ParseCommandLineOptions(argc, argv,
 534                               "llvm tool to verify use-list order\n");
 535
 536   LLVMContext Context;
 537   SMDiagnostic Err;
 538
 539   // Load the input module...
 540   std::unique_ptr<Module> M = parseIRFile(InputFilename, Err, Context);
 541
 542   if (!M.get()) {
 543     Err.print(argv[0], errs());
 544     return 1;
 545   }
 546   if (verifyModule(*M, &errs())) {
 547     errs() << argv[0] << ": " << InputFilename
 548            << ": error: input module is broken!\n";
 549     return 1;
 550   }
 551
 552   // Verify the use lists now and after reversing them.
 553   outs() << "*** verify-uselistorder ***\n";
 554   verifyUseListOrder(*M);
 555   outs() << "reverse\n";
 556   reverseUseLists(*M);
 557   verifyUseListOrder(*M);
 558
 559   for (unsigned I = 0, E = NumShuffles; I != E; ++I) {
 560     outs() << "\n";
 561
 562     // Shuffle with a different (deterministic) seed each time.
 563     outs() << "shuffle (" << I + 1 << " of " << E << ")\n";
 564     shuffleUseLists(*M, I);
 565
 566     // Verify again before and after reversing.
 567     verifyUseListOrder(*M);
 568     outs() << "reverse\n";
 569     reverseUseLists(*M);
 570     verifyUseListOrder(*M);
 571   }
 572
 573   return 0;
 574 }